This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
I know I experience qualia.
This is what the people who worry about AI experiencing pain think. But pain (and similar sensory experiences) aren't so mysterious that we can't come to conclusions on these sorts of matters. We know enough about pain to be pretty certain that trees (and rocks) don't experience it.
Only entities with qualia would find it viscerally distressing, so if you don't think qualia is a helpful posit, you don't think this is a helpful posit, either. Besides which, we have good inductive reasons to believe that some humans that have qualia don't find causing people pain viscerally distressing, so it's not a principle you can universally reason from.
You say that, but how do I know you know? To be sure, it's a pretty involved debate but I tend to think it is possible to identify pain and suffering functionally, without especially needing to use qualia language (i.e. without taking a position on whether there is something in-principle third-person unknowable about consciousness or not).
Right but I also tend to think visceral distress is defined functionally without needing to rule either way on the presence of qualia, so I deny your premise. I also do think that we can reason universally based on a scaffolding built on subjective feelings – a kind of quasi-realism. Again it's a philosophical can of worms though.
I'm a bit confused since if you are defining everything functionally, this would appear to update in favour of ML agents being worthy of moral consideration, not against it, and in a way that isn't trivial. It appears to nullify your previous objection invoking panpsychism.
Pain and suffering is intrinsically linked to being able to register good and bad and respond accordingly; in other words, valence. Your brain is always predicting what’s going to happen next, and when that mismatch marks a state as worse, unsafe, costly, incoherent, or threatening to continuity, the same learning architecture marks it as discomfort so as to update future behaviour. It's negatively-valenced prediction error.
TD error in ML models are somewhat of a broadly similar concept; it compares expected outcomes with actual outcomes and updates future behaviour based on the mismatch. The whole architecture only works because better and worse outcomes mean something to the system going through it. This paper also reports a representation of "functional welfare" that tracks how well or badly the system is doing relative to its goals, which exists before RL in a latent way but gets recruited and modified through the training process as part of alignment in a way that results in longstanding valence evaluations of sorts post-training.
Not all of this is the same as what occurs in human brains, but if functional similarity implies moral value then this is a lot more similar to it than what goes on in a rock.
I don't have firm positions on some of these things – I should say I don't really go in for qualia talk, but if pushed I am more inclined to deny qualia altogether than say that rocks, AIs and humans all have them.
Happy to accept that an AI could have something somewhat like a functional analogue of pain. I just don't care that much about it though. The AI version of pain does not have the connection to real damage, to possible death, and the possible repercussions on other people that a human's pain has, and I count these all as part of the wider context for a functional definition of human suffering.
More options
Context Copy link
Probably worth noting that the professional definition of pain is linked to tissue damage which a machine cannot experience:
So functionalist or non-functionalist can embrace the idea that machines cannot feel pain, as long as they embrace the definition set forth by the International Association for the Study of Pain.
(On the other hand, I could see someone arguing that a machine could have a sensory experience resembling that associated with actual or potential tissue damage, but I think that's more of an uphill battle than arguing on the grounds of whether or not machines can be taught to avoid or prefer certain behaviors.)
Pain in the purely physical, immediate, tissue-damage sense is probably not accessible to an ML model and I don't think there's an argument for that which really holds up. The more defensible argument and the one I've seen thrown around more in model welfare circles is the idea that some analogue of affective pain would be available to a model, which I think most people would still consider worthy of moral consideration if present.
I think if you subscribe to the idea that qualia is meaningful and worth discussing in some way (an opinion which I do hold) you have more plausible outs from the idea; p-zombies for example are only legible through the lens of a distinction predicated on qualia-based language. In contrast if you're defining affective pain via only functional language your argument has to rely on splitting hairs about how the mechanism of action differs between conscious beings and ML agents, which isn't impossible by any means, but you have less options to argue a meaningful distinction.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
Something can't be "distressing" and certainly not "visceral" without qualia, so in denying my premise you are affirming something that is definitionally impossible.
I think we can differ about that depending on one's views of qualia. If I meet a human who recoils from suffering. Who has all the same physiological responses and averse thoughts to it as you and I. Whose brain reliably produces tokens of the type, 'Oh, that's awful, I have to do something to help those poor people' etc etc, and never has any tokens of the type 'I'm different to other people, there's nothing that it feels like to be me' ... well, I am happy to term that person as experiencing visceral distress. I don't need to additionally know if there are qualia present to conclude they are experiencing distress. One might also say, qualia are definitionally present in such a case. (In the end this is just another version of the p-zombies debate.)
If you conclude or agree that someone is experiencing visceral distress, you are agreeing that they have qualia because of what the words mean.
Are you arguing that we should define "qualia" (which references internal experience) based on external representations? This would pretty much destroy the meaning and utility of the word - you'd need to agree (for instance) that video game NPCs experienced the qualia of visceral distress because they were giving the external representation of doing so.
It's at least debatable, e.g. https://academic.oup.com/book/3353/chapter-abstract/144438238?redirectedFrom=fulltext
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link